Methods and systems for filtering vehicle self-reflections in radar

Abstract

Example embodiments relate to self-reflection filtering techniques within radar data. A computing device may use radar data to determine a first radar representation that conveys information about surfaces in a vehicle's environment. The computing device may use a predefined model to generate a second radar representation that assigns predicted self-reflection values to respective locations of the environment based on the information about the surfaces conveyed by the first radar representation. The predefined model can enable a predefined self-reflection value to be assigned to a first location based on information about a surface positioned at a second location and a relationship between the first location and the second location. The computing device may then modify the first radar representation based on the predicted self-reflection values in the second radar representation and provide instructions to a control system of the vehicle based on modifying the first radar representation.

What is claimed is: 
     
       1. A method comprising:
 receiving, at a computing device and from a radar unit, radar data indicative of an environment; 
 based on the radar data, determining a first radar representation that conveys information about a plurality of surfaces in the environment; 
 generating, by the computing device and using a predefined model, a second radar representation that assigns predicted self-reflection values to respective locations of the environment based on the information about the plurality of surfaces conveyed by the first radar representation, wherein the predefined model enables a predefined self-reflection value to be assigned to a first location based on information about a surface positioned at a second location and a relationship between the first location and the second location; 
 modifying the first radar representation based on the predicted self-reflection values in the second radar representation; and 
 providing an alert based on modifying the first radar representation. 
 
     
     
       2. The method of  claim 1 , further comprising:
 causing, by the computing device, the radar unit to transmit radar signals having stretch linear frequency modulated (LFM) waveforms; and 
 wherein receiving radar data indicative of the environment comprises: 
 receiving radar data based on the radar signals having stretch LFM waveforms. 
 
     
     
       3. The method of  claim 1 , wherein the relationship between the first location and the second location indicates that the first location is positioned at a first range and a particular azimuth relative to the radar unit and the second location is positioned at a second range and the particular azimuth relative to the radar unit, and wherein the first range is approximately double the second range. 
     
     
       4. The method of  claim 1 , wherein determining the first radar representation that conveys information about the plurality of surfaces in the environment comprises:
 determining a radar data cube having a first plurality of voxels that each indicate information about a given surface in the environment. 
 
     
     
       5. The method of  claim 4 , wherein generating the second radar representation that assigns predicted self-reflection values to respective locations of the environment comprises:
 generating a reflection data cube having a second plurality of voxels that each represent a predicted self-reflection value for a given location of the environment using the predefined model, wherein the relationship between the first location and the second location further indicates that the first location includes a first Doppler relative to the radar unit and the second location includes a second Doppler relative to the radar unit, and wherein the first Doppler is approximately double the second Doppler. 
 
     
     
       6. The method of  claim 5 , wherein each voxel of the first plurality of voxels further indicates a range, an azimuth, and a Doppler for a given surface in the environment, and
 wherein each voxel of the second plurality of voxels further indicates a range, an azimuth, and a Doppler corresponding to a given location of the environment. 
 
     
     
       7. The method of  claim 6 , where modifying the first radar representation based on the predicted self-reflection values in the second radar representation comprises:
 performing a filter process that removes the reflection data cube from the radar data cube. 
 
     
     
       8. The method of  claim 7 , wherein the information about the plurality of surfaces indicated by the first plurality of voxels includes a radar cross section (RCS) measurement for each surface; and
 wherein performing the filter process that removes the reflection data cube from the radar data cube comprises: 
 subtracting the predicted self-reflection values represented by the second plurality of voxels from RCS measurements for the plurality of surfaces indicated by the first plurality of voxels. 
 
     
     
       9. The method of  claim 8 , further comprising:
 based on subtracting the predicted self-reflection values represented by the second plurality of voxels from RCS measurements for the plurality of surfaces indicated by the first plurality of voxels, performing object detection using the first plurality of voxels. 
 
     
     
       10. The method of  claim 1 , wherein determining the first radar representation that conveys information about the plurality of surfaces in the environment comprises:
 determining a two-dimensional radar image having a first plurality of pixels that each indicate information about a given surface in the environment. 
 
     
     
       11. The method of  claim 10 , wherein generating the second radar representation that assigns predicted self-reflection values to respective locations of the environment comprises:
 generating a two-dimensional reflection image having a second plurality of pixels that each represent a predicted self-reflection value to a given location of the environment. 
 
     
     
       12. The method of  claim 11 , wherein each pixel of the first plurality of pixels further indicates a range and an azimuth for a given surface in the environment, and
 wherein each pixel of the second plurality of pixels further indicates a range and an azimuth corresponding to a given location of the environment. 
 
     
     
       13. The method of  claim 12 , where modifying the first radar representation based on the predicted self-reflection values in the second radar representation comprises:
 performing a filter process that removes the two-dimensional reflection image from the two-dimensional radar image. 
 
     
     
       14. The method of  claim 13 , wherein the information about the plurality of surfaces indicated by the first plurality of pixels includes a radar cross section (RCS) measurement for each surface; and
 wherein performing the filter process that removes the two-dimensional reflection image from the two-dimensional radar image comprises: 
 subtracting the predicted self-reflection values represented by the second plurality of pixels from RCS measurements for the plurality of surfaces indicated by the first plurality of pixels. 
 
     
     
       15. The method of  claim 12 , wherein subtracting the predicted self-reflection values represented by the second plurality of pixels from RCS measurements for the plurality of surfaces indicated by the first plurality of pixels comprises:
 providing the two-dimensional reflection image and the two-dimensional radar image to a neural network that is configured to filter the predicted self-reflection values represented by the second plurality of pixels from RCS measurements for the plurality of surfaces indicated by the first plurality of pixels. 
 
     
     
       16. The method of  claim 1 , further comprising:
 based on modifying the first radar representation, identifying, using the first radar representation, one or more objects in the environment. 
 
     
     
       17. A system comprising:
 a radar unit; and 
 a computing device configured to:
 receive, from the radar unit, radar data indicative of an environment; 
 based on the radar data, determine a first radar representation that conveys information about a plurality of surfaces in the environment; 
 generate, using a predefined model, a second radar representation that assigns predicted self-reflection values to respective locations of the environment based on the information about the plurality of surfaces conveyed by the first radar representation, wherein the predefined model enables a predefined self-reflection value to be assigned to a first location based on information about a surface positioned at a second location and a relationship between the first location and the second location; 
 modify the first radar representation based on the predicted self-reflection values in the second radar representation; and 
 provide an alert based on modifying the first radar representation. 
 
 
     
     
       18. The system of  claim 17 ,
 wherein the relationship between the first location and the second location indicates that the first location is positioned at a first range and a particular azimuth relative to the radar unit and the second location is positioned at a second range and the particular azimuth relative to the radar unit, and 
 wherein the first range is approximately double the second range. 
 
     
     
       19. A non-transitory computer readable medium having stored therein instructions executable by one or more processors to cause a computing system to perform functions comprising:
 receiving, from a radar unit, radar data indicative of an environment; 
 based on the radar data, determining a first radar representation that conveys information about a plurality of surfaces in the environment; 
 generating, using a predefined model, a second radar representation that assigns predicted self-reflection values to respective locations of the environment based on the information about the plurality of surfaces conveyed by the first radar representation, wherein the predefined model enables a predefined self-reflection value to be assigned to a first location based on information about a surface positioned at a second location and a relationship between the first location and the second location; 
 modifying the first radar representation based on the predicted self-reflection values in the second radar representation; and 
 providing an alert based on modifying the first radar representation. 
 
     
     
       20. The method of  claim 1 , wherein the radar unit is coupled to a vehicle, and wherein providing the alert comprises:
 providing the alert via a display interface positioned inside the vehicle, wherein the display interface is viewable by one or more passengers located in the vehicle.